i5 2500k intel (oc to 4.5ghz)
asus gtx 460 768mb
1x4gb DDR-3 12800 1600mhz corsair vengeance ram (running at 1600)
No matter what I do, I am getting low 3d mark physics scores. 7097 physics score at 4.5ghz seems very low. I ran p95 for a few hours, everything was stable and temps hit 64C max. The score is low at all levels of overclocking, including stock. CPU-z shows my cpu is running at 4.5ghz and 800mhz for the ram (ddr would mean this is 1600). What is the problem? People are getting much higher scores at this speed.
Is the problem that I only have 1 ram chip? Does dual channel really give you that big of an increase in the physics score? I have searched everywhere and the only time others received low scores like this was with 2 or 3 sticks of ram running in single mode. C states did not change anything either.
Is there one specifically for CPUs that I could use to compare my system with others to see if there is just some odd problem with 3dmark scores? I have read that the physics test still has some dependency on the graphics card. I dont know if thats true though...
I still haven't been able to find anything related to the physics test comparison with 1 ram stick in single or 2 sticks in dual channel mode. Has anyone ever tried testing there system? I know that in other testing, the ram setup didnt make much of a difference but there are several examples on the web of people going from 2 sticks in single to 2 sticks in dual getting huge gains in 3dmark11. Maybe 1 stick in single vs 2 in dual would also give the same gain in performance? I am hesitant to buy another stick to try it out though...
The problem has to be because I am only using 1 ram stick. Has anyone ever compared 1 ram stick to two ram sticks using 3dmark11? I cant find anything like that on google. I can find other benchmarks or utilities but never 3dmark11.
I ordered a second (identical) ram stick. I guess I will see in two days...
I've never seen a comparison but it stands to reason that Dual Channel allows twice as much data to flow. I don't know how RAM intensive that benchmark is but certainly it seems plausible that single channel is causing your score to be off.
If you want to measure pure CPU performance maybe give Cinebench a try or Intel Burn Test, which gives a result in GFlops.